Insurance Thought Leadership reports that AI demand is driving unprecedented insurance risk for data centers. McKinsey projects nearly $7 trillion in global data-center investment by 2030, with more than 40% concentrated in the U.S. Swiss Re forecasts global insurance premiums tied to data centers will rise from $10.6 billion to $24.2 billion by 2030. The article covers growth across construction types - from hyperscale builds to core-and-shell conversions - and rising activity among colocation providers and enterprise operators. Principal exposures include business interruption and downtime, amplified by severe-weather siting; AM Best confirmed in June 2026 that more than 56% of planned U.S. data-center projects face major natural-catastrophe exposure. Carriers are responding with greater focus on loss control, engineering standards, and business continuity to contain revenue losses that can exceed physical damage costs.
What happened
Insurance Thought Leadership reports that AI-driven data-center expansion is reshaping insurance markets, citing McKinsey's projection of nearly $7 trillion in global data-center investment by 2030, with more than 40% of that capital concentrated in the U.S. The article documents growth across construction types - ground-up hyperscale builds, core-and-shell projects for later conversion, colocation expansion, and large enterprise-managed facilities - and examines the insurance implications of each. Swiss Re's sigma 07/2026 report projects global insurance premiums tied to data centers will more than double from $10.6 billion to $24.2 billion over the same period.
Key exposures (reported)
Business interruption and downtime is identified as the most consequential coverage area, given the scale of AI operations and the risk from complex power outages - power supply causes approximately 45% of data center outages. AM Best's June 2026 report flagged that more than half of planned U.S. data-center projects, representing roughly $670 billion in value, are in states at high risk of severe convective storms, tornadoes, and hail, and that 56% of planned facilities face at least one major natural catastrophe exposure. Additional gaps identified include off-premises power failure, delay-in-startup, advance loss of profits, pollution liability, and "silent cyber" coverage.
Market context
Average insured project values have surged from roughly $150 million to $3 billion in five years, per Risk & Insurance, straining traditional P/C market capacity. AM Best characterized the risk profile as beyond what the traditional property/casualty industry has previously experienced. Big-five cloud providers are forecast to spend more than $600 billion on capex in 2026, a 36% annual increase, with roughly 75% directly tied to physical AI infrastructure (Swiss Re sigma 07/2026).
Underwriting response
Carriers are placing greater emphasis on loss control, engineering standards, business continuity, and disaster recovery planning. The article reports that underwriting is shifting toward stricter redundancy requirements and resilience testing. Liquid-cooling-related losses now account for approximately 24% of total data-center loss costs, and lithium-ion battery integration into server racks has introduced ignition sources not previously present in traditional data processing environments.
What to watch
Practitioners should track:
- •how insurers incorporate grid resilience and supply-chain metrics into pricing models
- •whether policy forms evolve to address non-physical causes of outage
- •whether new instruments such as parametric covers or pooled catastrophe facilities emerge to manage correlated data-center loss exposures
Scoring Rationale #
Backed by McKinsey, Swiss Re sigma 07/2026, and AM Best, this is a substantive infrastructure risk story relevant to AI practitioners managing compute deployment and procurement. The $7 trillion capex projection and the doubling of insurance premiums to $24.2 billion by 2030 are material signals, but the primary angle is insurance-industry positioning rather than a direct AI development. Scores in the solid range: niche-relevant for infrastructure-aware practitioners, not broadly notable for AI/ML researchers.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems